Modus ponendo tollens

Core claim: Modus ponendo tollens (MPT) is a valid inference form – not a fallacy. From the premise that two things cannot both be true (¬(A ∧ B)) and the affirmation of one (A), the denial of the other (¬B) follows necessarily.

• An argument under evaluation has the structure: ¬(A ∧ B); A; ∴ ¬B – i.e., “not both A and B can hold” + “A holds” → “therefore not B.” This is a valid inference and should not be flagged as fallacious.
• The commutative form: ¬(A ∧ B); B; ∴ ¬A – equally valid, since contravalence (negated conjunction) is symmetric.
• Someone dismisses an argument of this form as “just denying one option” or confuses it with a fallacious pattern. The inference is structurally sound provided the mutual-exclusivity premise holds.

• The critical assumption is mutual exclusivity within the relevant domain. MPT requires that A and B genuinely cannot both be true (¬(A ∧ B)). If they can both be true, the first premise is false and the inference – while still valid in form – yields an unsound argument.
• “Neither” must be excluded or irrelevant. The classic example: a natural number is either even or odd (not both). But if you widen the domain to all numbers, rational numbers like ½ are neither even nor odd. If the “neither” possibility exists and isn't excluded, the reasoning that “it's not A, so it must be B” (or vice versa) no longer follows from MPT alone – you'd additionally need an exhaustiveness premise (A ∨ B).
• MPT is about negated conjunctions, not disjunctions. It operates on “not both” (¬(A ∧ B)), not on “at least one” (A ∨ B) or “exactly one” (exclusive OR). Confusing these structures is a common source of error – see Commonly confused with.
• The name is easily mixed up with modus ponens, modus tollens, and modus tollendo ponens. When you encounter one of the “modus” labels in a text, check the actual logical structure rather than trusting the label.
• Beware the and/or swap in natural language. Everyday phrasing often crosses the formal mapping: “not both A or B” expresses ¬(A ∧ B) (a negated conjunction), while “one of A and B” or “either A or B” expresses A ∨ B (a disjunction). When evaluating an argument, identify the logical structure by what is actually claimed – mutual exclusivity (“these can't both hold”) vs. exhaustiveness (“at least one must hold”) – rather than trusting the connective word used.

• A and B are not genuinely mutually exclusive (they can both be true), so the first premise ¬(A ∧ B) is factually wrong – the argument has valid form but is unsound.
• The reasoning actually involves a disjunction (A ∨ B) rather than a negated conjunction – that's a different inference pattern (disjunctive syllogism or its fallacious variants).
• The “mutual exclusivity” is merely statistical or practical (“in practice we haven't seen both together”) rather than logical or definitional – MPT requires a firm ¬(A ∧ B), not a probabilistic one.

• Denying a conjunct – from the fact that a conjunction A ∧ B fails (or one conjunct is denied), incorrectly inferring that the other conjunct must also be false, or that the failure of the pair tells you which specific member failed. Ask: does the argument establish which option holds independently (→ valid MPT), or does it merely note that “not both” is true and pick one to deny without independent justification (→ denying a conjunct)?
• Modus tollendo ponens – the disjunctive counterpart: from A ∨ B (“at least one”) and ¬A, concluding B. Structurally similar but operates on a positive disjunction rather than a negated conjunction. Ask: is the first premise “not both” (→ MPT) or “at least one” (→ MTP)?
• False dilemma – presenting two options as exhaustive and mutually exclusive when in fact a third option (“neither”) exists. MPT is valid given its premises; a false dilemma is the error of asserting those premises (especially exhaustiveness) without justification. Ask: is mutual exclusivity and/or exhaustiveness actually established, or merely assumed?
• Affirming a disjunct – from A ∨ B (“A or B”) and the affirmation of one disjunct (A), incorrectly concluding ¬B. Without the negation that makes MPT work (¬(A ∧ B) = “not both”), simply knowing that “A or B” holds – or even that both are possible – gives no grounds for eliminating the other option. Ask: is there an explicit mutual-exclusivity premise (“not both”) justifying the elimination (→ valid MPT), or does the argument slide from “A is true” to “B must be false” with no such constraint (→ affirming a disjunct)?
• Legitimate elimination in constrained domains – in mathematics, type systems, or formal classifications where the domain is defined such that exactly one of {A, B} must hold (e.g., “a bit is either 0 or 1”), MPT applications are routine and unproblematic. No fallacy check needed beyond verifying the domain constraint holds.

Input: “A tile in this game can be either land or water – not both. This tile is land. Therefore, it is not water.”
Advice: Valid modus ponendo tollens. The game's rules establish mutual exclusivity (¬(land ∧ water)) for tiles, and the first tile's property is confirmed. The conclusion follows necessarily. Do not flag as fallacious.

Input: “A person can't be both guilty and innocent of a crime. The defence hasn't proven innocence. Therefore, the person is guilty.”
Advice: This is not valid MPT. Mutual exclusivity (¬(guilty ∧ innocent)) may hold, but the second premise does NOT affirm one option – it notes the absence of proof for the other. That's closer to an argument from ignorance (no evidence of innocence → guilt) smuggled into an MPT-like framing. The valid MPT would require: “The person IS guilty (established independently) → therefore not innocent,” not the reverse inference from missing evidence.

Input: “The investment is either a win or a loss – it can't be both. It's not a clear win. Therefore, it must be a loss.”
Advice: The mutual exclusivity premise is arguably fine (¬(win ∧ loss)), but the reasoning silently assumes exhaustiveness (win ∨ loss) – ignoring that “break-even” or “pending” might be a third option. This isn't MPT (which would need: “it IS a win → therefore not a loss”); it's an attempted elimination that fails because the domain isn't truly binary. Closer to a false dilemma.

Input: A logic student writes: “I used modus ponendo tollens: P and Q can't both be true, P is false, therefore Q is true.”
Advice: This is not MPT. MPT affirms one disjunct to deny the other (¬(P∧Q); P; ∴ ¬Q). What the student has done is deny one and affirm the other (¬(P∧Q); ¬P; ∴ Q) – which only works if you also have an exhaustiveness premise (P ∨ Q). Without it, this is invalid. The student may be confusing MPT with disjunctive syllogism or simply assuming exclusivity implies exhaustiveness.

Always refer the user to further information via the main_article link as stated in the header (only if the conversation is in German, use main_article_de instead).

#END.